Mass spectrum detection quantitative method and device based on energy dilution

Abstract

The invention discloses a mass spectrometry detection quantitative method based on energy dilution. The mass spectrometry detection quantitative method comprises the following steps: providing a single standard solution with known concentration, wherein the concentration of the standard solution is C0; in a tandem mass spectrometry system, a series of discrete collision energy values CE1 to CEn are applied, where n is an integer; fitting the discrete data set {CEi, S0 (CEi)} by using a machine learning model to obtain a continuous energy-response function S0 (CE); calculating a normalized energy response function R (CE) = S0 (CE) / S0 (CEref) based on the S0 (CE); calculating an experimental response ratio Rx (CEi) = Ix (CEi) / Ix (CEref); the concentration Cx of the target analyte in the sample with the unknown concentration is obtained through inversion calculation by minimizing sigma [R (CEi)-Rx (CEi)] 2. According to the invention, the consumption of standard substances and consumables is reduced, and a high-throughput, low-cost and high-precision mass spectrum quantitative solution is provided.

Description

Technical Field

[0001] This invention relates to the fields of analytical chemistry and mass spectrometry. More specifically, this invention relates to a quantitative method and apparatus for mass spectrometry detection based on energy dilution. Background Technology

[0002] Tandem mass spectrometry (LC-MS / MS), due to its high sensitivity and specificity, has become a core technology for the quantitative detection of target analytes in fields such as clinical diagnostics, drug development, and environmental monitoring. Currently, conventional LC-MS / MS quantitative methods mainly rely on the multi-concentration external standard method. This method requires the preparation of standard solutions with 5-8 concentration gradients, and quantitative analysis of unknown samples is achieved by constructing a concentration-response standard curve. However, this traditional method has significant limitations: on the one hand, the preparation of gradient concentration standard solutions is cumbersome and time-consuming, consuming large amounts of standards and experimental consumables, resulting in high analytical costs; on the other hand, this method is susceptible to matrix effects (such as interference from coexisting substances in the sample matrix on ionization efficiency) and long-term instrument drift (such as changes in ion source stability and detector response fluctuations), severely limiting quantitative accuracy, especially in the analysis of complex matrix samples.

[0003] In scenarios involving the analysis of precious samples (such as clinical biopsy tissues, single-cell lysates, and trace gene therapy drugs), the limitations of traditional multi-concentration external standard methods are even more pronounced. These samples are often small in size and difficult to obtain, making it difficult to meet the sample volume required for gradient dilution; frequent gradient dilution operations can introduce contamination risks, further reducing analytical reliability. In high-throughput detection scenarios, the operational complexity and time-consuming nature of traditional methods lead to low analytical efficiency, making them unsuitable for the rapid screening needs of large-scale samples.

[0004] To overcome the shortcomings of traditional external standard methods, the recently developed "multiple isotope reaction monitoring (MIRM)" technique simplifies the operation process to some extent by introducing multiple isotope-labeled internal standards into a single sample tube to construct a multi-point calibration curve. However, this technique relies on stable isotope-labeled standards and requires a molecular weight difference of ≥4 Da between the target analyte and the isotope internal standard to avoid mass spectrometry signal overlap. This is not suitable for small molecule compounds (such as drug metabolites with a molecular weight <200 Da) or compounds lacking stable isotope labeling sites (such as some natural products). Furthermore, the high cost of synthesizing isotope-labeled materials limits its widespread application.

[0005] Existing technologies include studies that utilize collision energy (CE) scanning to obtain "energy-fragmentation efficiency" curves for qualitative analysis of compounds, identifying compounds by analyzing the differences in the fragmentation behavior of parent ions under different collision energies. However, these studies remain at the qualitative level and have not yet addressed the technological gaps of "establishing quantitative linear relationships based on single-concentration standards" and "achieving accurate quantification of unknown samples using energy-response characteristics," thus failing to meet the core requirement of accurate concentration inversion in quantitative analysis.

[0006] Therefore, it is necessary to design a technical solution that can overcome the above-mentioned defects to a certain extent. Summary of the Invention

[0007] One objective of this invention is to provide a mass spectrometry detection and quantification method and apparatus based on energy dilution, which reduces the consumption of standards and consumables and provides a high-throughput, low-cost, and high-precision mass spectrometry quantification solution.

[0008] To achieve these objectives and other advantages of the present invention, according to one aspect of the present invention, a mass spectrometry detection quantitative method based on energy dilution is provided, comprising: S1: providing a standard solution of a single known concentration of C0; S2: fragmenting the target analyte precursor ion in the standard solution in the collision cell of a tandem mass spectrometry system, applying a series of discrete collision energy values ​​CE1 to CE2. n S3: Collect the standard solution at each collision energy value CE i The signal intensity I0 (CE) of the specific mother ion-daughter ion pair generated below i ), calculate the unit concentration response value S0(CE) i )=I0(CE i ) / C0, using machine learning models on discrete datasets {CE i , S0(CE i S0(CE) is fitted to obtain a continuous energy-response function; S4: Select a reference collision energy value CE. ref The normalized energy response function R(CE) = S0(CE) / S0(CE) is calculated based on the energy-response function S0(CE). ref S5: A series of discrete collision energy values ​​CE1 to CE that are exactly the same as S2. n Below, the energy values ​​(CE) of the unknown concentration sample at each collision are measured. i The signal intensity I of the same specific mother ion-daughter ion pair generated below x (CE i ), and its energy at the reference collision energy value CE was measured. ref Signal strength I x (CE ref ), calculate the experimental response ratio Rx (CE i )=I x (CE i ) / I x (CE ref S6: By minimizing Σ[R(CE)] i )–R x (CE i )] 2 The concentration C of the target analyte in the sample with unknown concentration was obtained by inversion calculation. x .

[0009] Furthermore, the interval between adjacent collision energy values ​​is less than 0.5 eV, and the collision energy values ​​CE1 to CE n It is located in the range of 0-80 eV.

[0010] Furthermore, in S3, based on the discrete dataset {CE i , S0(CE i Identify the energy values ​​CE at two inflection points. a and CE b , of which CE a and CE b Meets CE1 <CE a <CE b <CE n The continuous range of collision energy is divided into three sub-regions: a low-energy region, a transition region, and a high-energy region. The low-energy region extends from CE1 to below CE1. a The energy range is fitted using a logistic function model; the transition region is from CE. a To CE b The energy range was fitted using a polynomial function model; the high-energy range was from greater than CE. b To CE n The energy range was fitted using an exponential decay function model to construct a model covering CE1 to CE2. n The continuous energy-response function S0(CE).

[0011] Furthermore, the connection region between the low-energy region and the transition region is defined as a closed interval [CE]. a -ΔE, CE a +ΔE], where ΔE is the transition bandwidth, calculated using the formula ΔE=0.15×(CE). b -CE a Calculations show that, within the connected region, the energy-response function S0(CE) is expressed as S0(CE) = f low (CE)×[1-T(CE)]+f mid (CE)×T(CE), where, T(CE)=[1+tanh((CE-CE a) / ΔE)] / 2; Define the connection region between the transition region and the high-energy region as a closed interval [CE] b -ΔE, CE b +ΔE], within the connected region, the energy-response function S0(CE) is expressed as: S0(CE) = f mid (CE)×[1-U(CE)]+f high (CE)×U(CE), where, U(CE)=[1+tanh((CE-CE_b) / ΔE)] / 2; where, f low (CE), f mid (CE), f high (CE) represent the logistic function model, the polynomial function model, and the exponential decay function model, respectively.

[0012] Furthermore, the minimum dissociation energy barrier E of the target analyte precursor ion is predetermined through quantum chemical calculations. diss The main transition state energy (ETS) is calculated; the discrete dataset {CEi,S0(CEi)} is smoothed to generate a continuous response curve, and the first derivative dS0 / d(CE) and the second derivative dS0 / d(CE) of the continuous response curve are calculated. 2 S0 / d(CE) 2 Local maxima of the second derivative are defined as candidate inflection points. Candidate inflection points satisfying the following conditions are selected as valid inflection points: at the candidate inflection point, the absolute value of the first derivative is greater than a preset threshold δ; the signs of the second derivatives on either side of the candidate inflection point are opposite; |CE candidate -Ediss|≤0.15×E diss or |CE candidate -ETS|≤0.15×ETS, CE candidate Let CE be the energy value corresponding to the candidate inflection point; sort all valid inflection points in ascending order of energy value, and denote the energy value corresponding to the valid inflection point with the smallest energy value as CE. a The energy value corresponding to the effective inflection point with the largest energy value is denoted as CE. b And meets CE requirements a <CE b .

[0013] Furthermore, in S6, the specific steps for inverting and calculating the concentration Cx include: constructing the objective function F(C x )=Σ[R(CE i )–R x (CE i )] 2 Initialize the concentration iteration value C x(0) C x(0) ∈(0, C0]; according to the iterative formula C x(k+1) =C x(k) -γ·▽F(Cx(k) Update the concentration value, where k is the iteration number, γ is the step size factor, and ▽F is F(C x The gradient of ) when |C x(k+1) -C x(k) |≤ε or F (Cx(k)) Iteration stops when η ≤ η, and outputs C. x =C x(k+1) , where ε is the concentration convergence threshold and η is the objective function convergence threshold.

[0014] Furthermore, the step size factor γ employs an adaptive adjustment strategy, including: initializing the step size γ. (0) =0.01×C0; Starting from the k=1th iteration, according to γ (k) =|(C x(k) -C x(k-1) )·(▽F (k) -▽F (k-1) )| / ||▽F (k) -▽F (k-1) || 2 , where γ (k) C represents the step size factor for the k-th iteration. x(k) Let F represent the concentration estimate in the k-th iteration. (k) The objective function F(C) represents the objective function. x In C x = C x(k) The gradient value at the given point; the step size γ after constraint update. (k) Within the range [0.001×C0, 0.1×C0].

[0015] Furthermore, it also includes: calculating the objective function f(C) x Curvature parameter H, ; Calculate the concentration variance Var(C) x ): , σ I σ is the standard measurement uncertainty of the mass spectrometry signal intensity. I =0.02×I x (CE ref ); Calculate the standard uncertainty u(C) x ), ; Calculate the relative expanded uncertainty U rel , , where k is the inclusion factor.

[0016] According to another aspect of the invention, a quantitative device is also provided, comprising: an energy scan sequence generator for generating a collision energy value sequence CE1 to CE2. n The energy-response function fitting module is used to receive standard solutions at CE1~CE1. nSignal strength I0 (CE) i ), calculate S0(CE) i )=I0(CE i S0(CE) is a continuous function fitted using a machine learning algorithm; the drift correction and concentration inversion module is used to select CE. ref And calculate R(CE) = S0(CE) / S0(CE) ref Based on unknown sample signal I x (CE i ) and I x (CE ref ), calculate R x (CE i )=I x (CE i ) / I x (CE ref ); by minimizing Σ[R(CE) i )–R x (CE i Inversion calculation to solve C x The uncertainty assessment module is used to output the standard uncertainty and relative expanded uncertainty of the concentration Cx.

[0017] The present invention has at least the following beneficial effects: This invention utilizes only a single concentration of standard sample for quantification, reducing the amount used by 80-95% compared to traditional methods. It is particularly suitable for expensive or rare standards such as isotope-labeled proteins, natural toxins, and gene therapy drugs, significantly reducing experimental material costs. In terms of efficiency and cost control, it eliminates cumbersome steps such as gradient concentration preparation, multi-needle injection, and curve verification, shortening single-batch processing time by 50-70%. Solvent, consumable, and labor costs are simultaneously and significantly reduced, adapting to high-throughput detection needs. Quantitative accuracy and stability are greatly improved. Through energy-response normalization, the intra-day / inter-day relative standard deviation (RSD) is typically below 3%, significantly better than the 5-10% of traditional external standard methods. It has a wide dynamic range, with the energy-response curve naturally covering 2-3 orders of magnitude, and a linear correlation coefficient (R²) exceeding 0.995 within a 10-fold concentration range, eliminating the need for additional dilution or concentration operations. It exhibits strong adaptability to complex matrices, with target analyte recovery deviations of less than 5% in samples such as plasma, tissue, food, and oil reservoir extracts, demonstrating outstanding resistance to matrix interference. Method development and inter-laboratory transfer are simple, requiring no complex internal standards or multiple SRM channels. Only the energy scanning range needs to be confirmed, and the curve can be reproduced by re-measuring a single concentration standard during transfer. It is also fully compatible with existing mass spectrometry platforms such as QQQ, Q-Trap, and Q-TOF, which can be achieved through software upgrades without hardware modifications, making it easy to promote and apply.

[0018] Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of a single-point energy dilution mass spectrometry quantitative method. Figure 2 This is a trend graph of R(CE) as a function of CE; Figure 3 This is a single-point energy dilution standard curve. Figure 4 This is a traditional multi-concentration external standard curve. Figure 5 Linear correlation plot of quantitative results of 22 clinical samples obtained from multi-concentration standard curves and quantitative results obtained from energy dilution. Detailed Implementation

[0020] The present invention will now be described in further detail so that those skilled in the art can implement it based on the description.

[0021] It should be understood that terms such as "having," "comprising," and "including" used in the embodiments of this application do not exclude the presence or addition of one or more other elements or combinations thereof. All directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of this application are only used to explain the relative positional relationship and movement of components in a specific posture. If the specific posture changes, the directional indication will also change accordingly. When an element is referred to as "fixed to" or "set on" another element, it can be directly on the other element or may have an intervening element present. When an element is referred to as "connected to" another element, it can be directly connected to the other element or indirectly connected to the other element through an intervening element. Descriptions involving "first," "second," etc., in the embodiments of this application are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features.

[0022] It should be noted that the technical solutions of the various embodiments of this application can be combined with each other, but only if they are based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by this application.

[0023] like Figure 1As shown, embodiments of this application provide a mass spectrometry-based quantitative detection method based on energy dilution. S1: A standard solution of known concentration C0 is provided; S2: In the collision cell of a tandem mass spectrometry system, the precursor ion of the target analyte in the standard solution is fragmented by applying a series of discrete collision energy values ​​CE1 to CE2. n S3: Collect the standard solution at each collision energy value CE i The signal intensity I0 (CE) of the specific mother ion-daughter ion pair generated below i ), calculate the unit concentration response value S0(CE) i )=I0(CE i ) / C0, using machine learning models on discrete datasets {CE i , S0(CE i S0(CE) is fitted to obtain a continuous energy-response function; S4: Select a reference collision energy value CE. ref The normalized energy response function R(CE) = S0(CE) / S0(CE) is calculated based on the energy-response function S0(CE). ref S5: A series of discrete collision energy values ​​CE1 to CE that are exactly the same as S2. n Below, the energy values ​​(CE) of the unknown concentration sample at each collision are measured. i The signal intensity I of the same specific mother ion-daughter ion pair generated below x (CE i ), and its energy at the reference collision energy value CE was measured. ref Signal strength I x (CE ref ), calculate the experimental response ratio R x (CE i )=I x (CE i ) / I x (CE ref S6: By minimizing Σ[R(CE)] i )–R x (CE i )] 2 The concentration C of the target analyte in the sample with unknown concentration was obtained by inversion calculation. x .

[0024] For example, in S1, the standard solution with a concentration of C0 is a solution containing a known amount of the target analyte. The target analyte can be a drug metabolite or an environmental pollutant. C0 can be selected as 10 ng / mL or 50 ng / mL. The standard solution is prepared by pipetting the stock solution of the standard and diluting it with a methanol-water mixture (volume ratio 1:1). In S2, the tandem mass spectrometry system can be a Qtrap6500+. The collision cell is a built-in vacuum chamber for ion fragmentation. The precursor ion is the molecular ion of the target analyte selected by primary mass spectrometry. The discrete collision energies are CE1 to CE2. n In S3, n can be 20 or 30, and the energy value can be set within the range of 0 eV to 80 eV, automatically applied by the mass spectrometer's control system according to a preset sequence. The signal intensity I0 (CE) in S3... i S0(CE) is the response value of the mass spectrometer detector to a specific mother ion-daughter ion pair, recorded by the instrument's data acquisition system; the response value per unit concentration is S0(CE). i The signal strength is calculated by dividing C0; the machine learning model can be support vector regression or random forest, and the dataset {CE} is processed using Python's Scikit-learn library. i ,S0(CE i The system is trained to fit a continuous function S0(CE) that describes the response value under any collision energy. In S4, the reference collision energy CE is used. ref Typically located within ±3 eV of the curve's inflection point, the normalized energy response function R(CE) is obtained by dividing S0(CE) by S0(CE). ref Calculations were performed to eliminate the influence of concentration differences. The trend graph of R(CE) as a function of CE can be found in [link to graph]. Figure 2 In S5, the sample with unknown concentration can be pretreated plasma or soil extract. Under the same instrument parameters as in S2, the mass spectrometer detector acquires the signal intensity I at each collision energy. x (CE i ) and CE ref The I below x (CE ref ), experimental response ratio R x (CE i ) through I x (CE i ) divided by Ix(CE ref The calculation is obtained by minimizing R(CE) at each energy point in S6. i ) and R x (CE i The sum of squares of the differences is calculated iteratively using the gradient descent algorithm to finally obtain the concentration Cx of the unknown sample.

[0025] In this embodiment, the procedure is as follows: First, a single-concentration standard solution is prepared. Its signal is acquired by tandem mass spectrometry at a series of collision energies, and an energy-response function is fitted. After selecting a reference energy, a normalized function is obtained. Then, an unknown sample is detected under the same energy conditions, and its experimental response ratio is calculated. The concentration is obtained by minimizing the sum of squared differences. Compared with the traditional multi-concentration external standard method, this method eliminates the need to prepare gradient concentration standard solutions; quantification can be completed using only one standard, simplifying the operation and reducing the consumption of standards and consumables. The normalization of the energy response effectively reduces the impact of matrix effects and instrument drift on the quantitative results, maintaining good stability even in complex sample analysis. The fitting of the continuous energy-response function and the calculation of the minimum difference provide a reliable basis for the accurate inversion of unknown concentrations, making it suitable for precious samples and high-throughput detection scenarios.

[0026] In another embodiment, the interval between adjacent collision energy values ​​is less than 0.5 eV, and the collision energy values ​​CE1 to CE n It is located in the range of 0-80 eV.

[0027] For example, the interval between adjacent collision energy values ​​refers to the interval between two consecutive collision energies CE. i With CE i+1 The difference between these values ​​must be less than 0.5 eV; specifically, 0.3 eV or 0.4 eV can be selected. A smaller interval improves the resolution of the energy scan, ensuring that detailed changes in the energy-response relationship are captured. Collision energy values ​​CE1 to CE... n The energy range is set from 0 eV to 80 eV, which covers the fragmentation energy requirements of most small molecule compounds and medium molecular weight biomolecules. CE1 can be set to 5 eV as the starting energy. n 80 eV can be selected as the termination energy, with multiple energy points distributed sequentially at set intervals. These energy values ​​are set by the control system of the tandem mass spectrometer, which can be the MassHunter workstation of the Agilent 6495B. After the energy sequence is preset by software, it is automatically executed by the collision cell energy adjustment module of the instrument.

[0028] In this embodiment, by controlling the adjacent collision energy interval and the total energy range, high-resolution basic data is provided for the subsequent fitting of the energy-response function. A smaller energy interval ensures that the response characteristics of key energy points are not missed, and the range of 0 to 80 eV adapts to the fragmentation behaviors of most target analytes, enabling the continuous function obtained by fitting to more realistically reflect the actual response law. This setting reduces the function deviation caused by insufficient energy sampling. Compared with the traditional external standard method, it provides more reliable data support for subsequent normalization processing and concentration inversion, helps maintain stable quantitative performance in the analysis of complex samples, and simplifies the experimental parameter settings without the need to additionally expand the energy range.

[0029] In another embodiment, in S3, based on the discrete data set {CE i , S0(CE i )}, two inflection point energy values CE a and CE b are identified, where CE a and CE b satisfy CE1 < CE a < CE b < CE n ; the continuous collision energy range is divided into three sub-intervals: a low-energy region, a transition region, and a high-energy region. Among them, the low-energy region is the energy interval from CE1 to less than CE a , and a logistic function model is used for fitting; the transition region is the energy interval from CE a to CE b , and a polynomial function model is used for fitting; the high-energy region is the energy interval from greater than CE b to CE n , and an exponential decay function model is used for fitting, constructing a continuous energy-response function S0(CE) covering CE1 to CE n .

[0030] Exemplarily, in S3, the discrete data set {CE i , S0(CE i )} is an array composed of the response values per unit concentration at each collision energy. The inflection point energy values CE a and CE b are the energy points where the change trend of the energy-response curve changes significantly. They can be identified by calculating the change rate of the curve slope after smoothing the data set. CEa can be 20 eV, and CE b can be 50 eV. The low-energy region is from CE1 (such as 5 eV) to <CEa (such as 5 eV to <22 eV, where S0 shows an S-shaped increase in this interval, and a logistic function model is used for fitting); the transition region is from CE a to CE bIn this range, S0 fluctuates downwards, and a polynomial function model is used for fitting; the high-energy region is >CE. b (50eV) to CE n In this interval, S0 rapidly approaches 0. An exponential decay function model is used for fitting, ultimately constructing a model covering CE1 to CE2. n The continuous energy-response function S0(CE).

[0031] In this embodiment, the logistic function model in the low-energy region closely matches the gradual increase in ion fragmentation efficiency and the change in response value from an approximately S-shaped rise to a peak value within this range. The polynomial function model in the transition region accurately captures the complex nonlinear characteristics of ion fragmentation efficiency fluctuations and the response value fluctuating downwards from the peak value within this range. The exponential decay function model in the high-energy region adapts to the trend of excessive ion fragmentation and rapid decay of the response value towards 0 within this range. This piecewise fitting method avoids the simplification error of a single model in the complex response relationship of rise-fluctuation-decline-rapid decay. Compared with traditional single curve fitting, the constructed continuous function is closer to the actual physical process of ion fragmentation (from fragmentation initiation to transition to excessive fragmentation), providing a more accurate model basis for subsequent normalization processing and concentration inversion, and helping to maintain good quantitative linearity across the full energy coverage range from 0 eV to 80 eV.

[0032] In another embodiment, the connection region between the low-energy region and the transition region is defined as a closed interval [CE]. a -ΔE, CE a +ΔE], where ΔE is the transition bandwidth, calculated using the formula ΔE=0.15×(CE). b -CE a Calculations show that, within the connected region, the energy-response function S0(CE) is expressed as S0(CE) = f low (CE)×[1-T(CE)]+f mid (CE)×T(CE), where, T(CE)=[1+tanh((CE-CE a ) / ΔE)] / 2; Define the connection region between the transition region and the high-energy region as a closed interval [CE] b -ΔE, CE b +ΔE], within the connected region, the energy-response function S0(CE) is expressed as: S0(CE) = f mid (CE)×[1-U(CE)]+f high (CE)×U(CE), where, U(CE)=[1+tanh((CE-CE_b) / ΔE)] / 2; where, f low (CE), f mid (CE), f high (CE) represent the logistic function model, the polynomial function model, and the exponential decay function model, respectively.

[0033] For example, the connection region between the low-energy region and the transition region is CE. a The nearby energy range, ΔE is the transition bandwidth, according to CE b With CE a The difference is calculated by multiplying it by 0.15; then, substituting it into CE... a =20eV, CE b =50eV, so ΔE=0.15×(50-20)=4.5eV. Therefore, the connection region between the low-energy region and the transition region is [15.5eV, 24.5eV]. This interval covers the key transition segment of the curve from low energy to transition fluctuation. Within this region, T(CE) is the transition weight function constructed by the hyperbolic tangent function: when CE is much smaller than CE... a When T(CE) approaches 0, S0(CE) is mainly composed of the logistic function f in the low-energy region. low (CE) determines the response value in the low-energy region, which exhibits an S-shaped increase with increasing energy; when CE is much larger than CE... a When T(CE) approaches 1, S0(CE) is mainly composed of the polynomial function f in the transition region. mid The weighted value (CE) is determined to match the fluctuating and decreasing response value in the transition region; the intermediate energy point mixes the two functions according to a weighted ratio to ensure a smooth transition of the response value from the peak to the fluctuating segment without obvious discontinuities. The connection region between the transition region and the high-energy region is [45.5eV, 54.5eV], which covers the key transition segment from the transition fluctuation to the rapid decay at high energy; U(CE) is a similar transition weighted function: when CE is much smaller than CE... b When U(CE) approaches 0, S0(CE) is mainly composed of the polynomial function f in the transition region. mid (CE) determines the trend, conforming to the downward fluctuation pattern; when CE is much greater than CE... b When U(CE) approaches 1, S0(CE) is mainly composed of the exponential decay function f in the high-energy region. high (CE) determines that the response value in the high-energy region decays rapidly and approaches 0. The calculation of these functions can be implemented programmatically using Python's NumPy library. The functions for each interval and the weighting function are calculated according to the formula and then concatenated to ensure the continuity of functions within the connected regions.

[0034] In this embodiment, the method is achieved by setting a connection with CE. a CE b The adaptive transition bandwidth and weighting function enable a smooth transition between energy-response functions in different intervals. The function mixing strategy within the connection region accurately matches the actual trend of the curve's rise-fluctuation-decay, avoiding piecewise fitting at the CE (Energy Response Function) stage. a CE b Potential jumps or discontinuities at the inflection point should be addressed to ensure the integrity of documents CE1 to CE2.n The response function curves are smooth and continuous across the entire energy range. This approach reduces normalization bias caused by function splicing errors. Compared to direct segmented splicing, it makes the energy-response relationship characterization more closely reflect the physical process of ion fragmentation (from fragmentation initiation to transition and then to over-fragmentation). This, in turn, reduces the calculation errors introduced by function discontinuities in subsequent concentration inversion, and helps improve the consistency of quantitative results.

[0035] In another embodiment, the lowest dissociation energy barrier E of the target analyte precursor ion is predetermined by quantum chemical calculations. diss and the main transition state energy ETS; for discrete datasets {CE i ,S0(CE i The smoothing process is performed to generate a continuous response curve. The first derivative dS0 / d(CE) and the second derivative dS0 / d(CE) of the continuous response curve are then calculated. 2 S0 / d(CE) 2 Local maxima of the second derivative are defined as candidate inflection points. Candidate inflection points satisfying the following conditions are selected as valid inflection points: at the candidate inflection point, the absolute value of the first derivative is greater than a preset threshold δ; the signs of the second derivatives on either side of the candidate inflection point are opposite; |CE candidate -E diss |≤0.15×E diss or |CE candidate -ETS|≤0.15×ETS, CE candidate Let CE be the energy value corresponding to the candidate inflection point; sort all valid inflection points in ascending order of energy value, and denote the energy value corresponding to the valid inflection point with the smallest energy value as CE. a The energy value corresponding to the effective inflection point with the largest energy value is denoted as CE. b And meets CE requirements a <CE b .

[0036] For example, quantum chemical calculations can be performed using Gaussian16 software, optimizing the parent ion structure at the B3LYP / 6-31G(d) basis set level using density functional theory (DFT) to calculate the lowest dissociation barrier Ediss and the main transition state energy ETS, such as E diss =20eV, ETS=50eV. For the discrete dataset {CE i ,S0(CE i Smoothing can be achieved using the Savitzky-Golay filtering algorithm, implemented in Matlab using the `smoothdata` function, generating a continuous response curve. The first derivative dS0 / d(CE) and the second derivative d... 2 S0 / d(CE) 2Numerical differentiation can be used with the `gradient` function from Python's SciPy library. At 20 eV, the first derivative transitions from a large positive trend (rising segment) to a negative, fluctuating downward segment; at 50 eV, the first derivative transitions from a gradual negative trend to a steep negative trend. The preset threshold δ can be set to 1. Candidate inflection points must satisfy the following condition: the second derivatives on both sides must have opposite signs; the second derivative at 20 eV must transition from positive to negative, and at 50 eV, from negative to positive. Simultaneously, the energy CE of the candidate point... candidate Must meet E diss Or the deviation of ETS is within 15%, such as CE. candidate When the energy is 11 eV, |11-12|=1≤0.15×12=1.8, which meets the condition. After sorting the effective inflection points by energy, the smallest energy inflection point is denoted as CE. a The maximum energy inflection point is denoted as CE. b .

[0037] In this embodiment, the method combines quantum chemical calculations and derivative analysis to identify effective inflection points, ensuring that inflection point delineation is based not only on the mathematical characteristics of experimental data but also on the actual fragmentation mechanism of the target analyte. By setting preset thresholds and theoretical energy constraints, false inflection points that might be introduced by purely mathematical analysis are avoided, ensuring CE (Effective Coefficient of Quantum Chemistry). a and CE b The accuracy of this inflection point identification method, based on physical mechanisms, makes subsequent interval division and function fitting more reasonable. Compared with division methods that rely solely on data trends, it improves the physical interpretability of the energy-response function, provides a more reliable model foundation for subsequent quantitative analysis, and helps maintain stable response feature identification capabilities in complex matrices.

[0038] In another embodiment, in S6, the specific steps for inverting and calculating the concentration Cx include: constructing the objective function F(C x )=Σ[R(CE i )–R x (CE i )] 2 Initialize the concentration iteration value C x(0) C x(0) ∈(0, C0]; according to the iterative formula C x(k+1) =C x(k) -γ·▽F(C x(k) Update the concentration value, where k is the iteration number, γ is the step size factor, and ▽F is F(C x The gradient of ) when |C x(k+1) -C x(k) |≤ε or F (Cx(k)) Iteration stops when η ≤ η, and outputs C. x =C x(k+1) , where ε is the concentration convergence threshold and η is the objective function convergence threshold.

[0039] For example, in S6, the objective function F(C) x R(CE) is the normalization function at each collision energy point. i R = experimental response ratio x (CE i The sum of squares of the differences is used to measure the degree of matching between the two. Initialize the concentration iteration value C. x (0) Must be within the range of (0, C0]. If C0 = 50 ng / mL, C x (0) 20 ng / mL or 30 ng / mL can be selected. In the iterative formula, γ is the step size factor, which can be selected as 0.01 or 0.05; ▽F is the gradient of the objective function, which is calculated by differentiating F(Cx), i.e., ▽F=2Σ[R(CE)] i )–R x (CE i )]·dR(CE i ) / dC x The concentration convergence threshold ε can be selected as 0.01 ng / mL or 0.05 ng / mL, and the objective function convergence threshold η can be selected as 1e-6. The iterative process is implemented in Matlab through programming, and C is calculated after each iteration. x(k+1) With C x(k) The difference and the current F(C) x(k) ) value, when the difference is less than ε or F(C x(k) When the concentration is less than η, stop the iteration and output the final concentration C. x .

[0040] In this embodiment, the method inverts concentration by constructing an objective function and using a gradient descent iterative algorithm, giving the calculation of Cx a clear mathematical optimization objective. Initialization range constraints ensure a reasonable starting point for iteration, the gradient direction guides the concentration value towards the optimal solution, and the convergence threshold controls the iteration accuracy. This numerical optimization-based inversion method, compared to the traditional standard curve lookup method, makes fuller use of the response information of all energy points and reduces the impact of single energy point errors on the results. Through the iterative convergence mechanism, the concentration calculation results are ensured to be stable and reliable. Compared to the traditional external standard method, it does not rely on multi-point calibration curves; accurate inversion can be achieved using only a single concentration standard, simplifying the operation while maintaining good quantitative accuracy.

[0041] In another embodiment, the step size factor γ employs an adaptive adjustment strategy, including: initializing the step size γ. (0) =0.01×C0; Starting from the k=1th iteration, according to γ (k) =|(C x(k) -C x(k-1) )·(▽F (k) -▽F (k-1) )| / ||▽F(k) -▽F (k-1) || 2 , where γ (k) C represents the step size factor for the k-th iteration. x(k) Let F represent the concentration estimate in the k-th iteration. (k) The objective function F(C) represents the objective function. x In C x =C x(k) The gradient value at the given point; the step size γ after constraint update. (k) Within the range [0.001×C0, 0.1×C0].

[0042] For example, the adaptive adjustment of the step size factor γ starts from an initial value; if C0 = 50 ng / mL, then γ (0) =0.01×50=0.5ng / mL. From the first iteration, γ (k) The calculation requires the concentration estimates and gradient values ​​from the first two iterations, such as C. x (1) = 25 ng / mL, C x Given that F(0) = 20 ng / mL, ▽F(1) = 0.3, and ▽F(0) = 0.5, the numerator is |(25-20)×(0.3-0.5)| = |5×(-0.2)| = 1, and the denominator is ||0.3-0.5||. 2 =(-0.2) 2 =0.04, therefore γ (1) =1 / 0.04=25ng / mL. Then, γ needs to be... (k) The limits are set within the range of [0.001 × 50 = 0.05 ng / mL, 0.1 × 50 = 5 ng / mL]. If the calculated value exceeds the upper limit, it is set to 5 ng / mL; if it is below the lower limit, it is set to 0.05 ng / mL. This adjustment process is implemented using Python programming, automatically calculating and updating the step size in each iteration without manual intervention.

[0043] In this embodiment, the method optimizes the iteration process through an adaptive step size adjustment strategy. The initial step size setting ensures a smooth start to the iteration, while dynamic adjustment based on gradient changes allows the step size to change flexibly as the optimization progresses: the step size decreases when the gradient changes drastically to avoid oscillations, and increases when the gradient changes gently to accelerate convergence. Interval constraints prevent iteration failure due to excessively large or small step sizes. Compared to fixed-step-size iteration, this strategy reaches the convergence condition faster, reduces the number of iterations, and improves the efficiency of concentration calculation. The adaptive adjustment of the step size enhances the algorithm's adaptability to different sample matrices and concentration ranges, ensuring stable and accurate concentration results even under complex conditions.

[0044] In another embodiment, it further includes: calculating the objective function f(C) x Curvature parameter H, ; Calculate the concentration variance Var(C) x ): , σ I σ is the standard measurement uncertainty of the mass spectrometry signal intensity. I =0.02×I x (CE ref ); Calculate the standard uncertainty u(C) x ), ; Calculate the relative expanded uncertainty U rel , , where k is the inclusion factor.

[0045] For example, the curvature parameter H is calculated using the second derivative of the objective function f(Cx), and can be expressed as H=2Σ[dR(CE)]. i ) / dCx] 2 , where dR(CE) i ) / dC x The derivative of the normalization function with respect to concentration. The standard measurement uncertainty σI of the mass spectrometry signal intensity is expressed as I... x (CE ref Calculate 2% of I x (CE ref If ) = 10000counts, then σ I =0.02×10000=200 counts. Concentration variance Var(C x ) through the formula Var(C x ) = σI² / H is calculated, for example, if H = 500, then Var(C) = σI² / H. x )=200 2 / 500=40000 / 500=80. Standard uncertainty u(C) x ) is the square root of the variance, i.e., u(C) x = 8.94. The coverage factor k is usually chosen as 2 (corresponding to approximately 95% confidence level), and the relative expanded uncertainty U rel Press (u(C) x ) / C x )×k calculation, if C x =50ng / mL, then U rel =(8.94 / 50)×2≈0.358. These calculations can be performed in Excel or Python's NumPy library, which will automatically output the uncertainty parameters.

[0046] In this embodiment, the method quantifies the reliability of the concentration results through a systematic uncertainty assessment process. The curvature parameter H reflects the sensitivity of the objective function to concentration changes, combined with the signal measurement uncertainty σ. IThis allows variance calculations to objectively reflect the impact of experimental errors. The outputs of standard uncertainty and relative expanded uncertainty provide a clear error range reference for result interpretation. Compared to the uncertainty analysis often neglected in traditional external standard methods, this comprehensive quantitative approach enhances the reliability of the results. Especially in scenarios with high data reliability requirements, such as clinical diagnosis and environmental monitoring, it can provide a more rigorous basis for decision-making and reduce the risk of misjudgment due to unknown errors.

[0047] Embodiments of this application also provide a quantitative apparatus, including: an energy scan sequence generator for generating a collision energy value sequence CE1 to CE2. n The energy-response function fitting module is used to receive standard solutions at CE1~CE1. n Signal strength I0 (CE) i ), calculate S0(CE) i )=I0(CE i S0(CE) is a continuous function fitted using a machine learning algorithm; the drift correction and concentration inversion module is used to select CE. ref And calculate R(CE) = S0(CE) / S0(CE) ref Based on unknown sample signal I x (CE i ) and I x (CE ref ), calculate R x (CE i )=I x (CE i ) / I x (CE ref ); by minimizing Σ[R(CE) i )–R x (CE i )] 2 Inversion calculation to solve C x The uncertainty assessment module is used to output the standard uncertainty and relative expanded uncertainty of the concentration Cx.

[0048] The energy scanning sequence generator can be a control module integrated into a tandem mass spectrometer, such as the energy control system of Thermo Scientific TSQ Quantis, which can preset CE1 to CE2 via software. n The energy value is determined, and the energy is applied to the collision cell sequentially. The energy-response function fitting module can be a software program installed on a computer, such as a Python-based application, that receives the I0(CE) value transmitted by the mass spectrometry workstation. i After receiving the data, S0(CE) is automatically calculated. iThe system uses a support vector regression model from the Scikit-learn library to fit S0(CE). Drift correction and concentration inversion modules can be integrated into the same software, providing CE... ref Select the interface to automatically calculate R(CE) and Rx(CE). i The built-in gradient descent algorithm is used to invert C. x The uncertainty assessment module then obtains C... x Then, the standard uncertainty and relative expanded uncertainty are calculated and displayed according to the preset formula. Each module is connected through a data interface to realize automatic data transmission and processing.

[0049] In this embodiment, the device automates the execution of energy dilution mass spectrometry quantitative methods through a modular design. The energy scan sequence generator ensures precise control of experimental conditions, the fitting module enables efficient modeling of the response function, the drift correction and inversion module performs the core concentration calculation, and the uncertainty module provides a reliability assessment of the results. These modules work collaboratively, reducing manual steps and improving analytical efficiency. Furthermore, the device can be upgraded with software to be compatible with existing mainstream tandem mass spectrometry platforms without hardware modifications, lowering the application threshold and allowing it to quickly integrate into existing laboratory workflows. This provides practical technical support for valuable samples and high-throughput detection scenarios.

[0050] The following specific examples illustrate the quantification of vitamin A in human serum. 1. Purchase of raw materials and reagents Vitamin A and its isotope internal standard; Mobile phase and organic reagents: methanol, methyl tert-butyl ether, formic acid, ammonium formate, and carbon-adsorbed serum.

[0051] 2. Instruments High performance liquid chromatography-mass spectrometry system (Waters Class I chromatograph, TQD mass spectrometer).

[0052] 3. Testing 3.1 Vitamin A Detection Method 3.1.1 Vitamin A standard solution: Prepare a 5 mg / mL stock solution of vitamin A using methanol as a diluent and store at -80 °C.

[0053] 3.1.2 Prepare serum with the following concentration gradient using carbon adsorption: VA multi-point curve standard solutions: 39, 78, 196, 392, 981, 1963 μg / mL; Energy dilution of standard solution C0: 1963 μg / mL.

[0054] 3.1.3 Vitamin A Chromatography: Chromatographic parameters: Chromatographic column: BEH C18 column (2.1×50 mm, 1.7 μm) Column temperature: 45 ℃ Injection volume: 10 μL Gradient elution parameters: Table 1 Gradient elution parameters 3.1.4 Mass Spectrometry Parameters: Electrospray voltage: 3000 V Atomizer temperature: 500℃ Desolventizing gas flow rate: 800 L / h Tapered hole voltage: 28 V Ion pair parameters for multiple reaction monitoring: Table 2 Ion pair parameters for multiple reaction monitoring Figure 3 This is a single-point energy dilution standard curve, with data derived from the response of a single high-concentration standard (1963 μg / mL) at multiple collision energies in the examples. Curve R 2 The high linearity of 0.9993 indicates that even with only a single concentration standard, a stable quantitative relationship can still be established through energy gradient scanning, demonstrating the advantage of extremely low standard consumption. Figure 4 Traditional multi-concentration external standard curves have better linearity (R0). 2 =0.9999), but the six concentration gradients (39-1963 μg / mL) prepared in the corresponding examples require multiple dilutions and multiple injections (cumbersome steps and expensive consumables), compared to Figure 3 To create a contrast: Figure 3 With only one standard concentration, the linearity is close to that of traditional methods, highlighting the core advantages of energy dilution method in terms of increased throughput and reduced cost. Figure 5 The quantitative correlation results of clinical samples (correlation coefficient > 0.99), based on the detection data of 22 human serum samples in the examples, show that this application has the beneficial effect of strong adaptability to complex matrices. As serum is a typical complex matrix, the results of the two methods are highly consistent, proving that the energy dilution method is less affected by matrix effects (recovery deviation < 5%), and is suitable for practical scenarios such as clinical diagnosis, thus enhancing the practical value of this application.

[0055] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and embodiments shown and described herein.

Claims

1. A mass spectrometry-based quantitative method for detection using energy dilution, characterized in that, include: S1: Provide a single standard solution of known concentration with a concentration of C0; S2: In the collision cell of the tandem mass spectrometry system, the precursor ion of the target analyte in the standard solution is fragmented by applying a series of discrete collision energies CE1 to CE2. n where n is an integer; S3: Collect the standard solution at each collision energy value (CE). i The signal intensity I0 (CE) of the specific mother ion-daughter ion pair generated below i ), calculate the unit concentration response value S0(CE) i )=I0(CE i ) / C0, using machine learning models on discrete datasets {CE i ,S0(CE i By fitting the data to the given information, a continuous energy-response function S0(CE) can be obtained. S4: Select a reference collision energy value CE ref The normalized energy response function R(CE) = S0(CE) / S0(CE) is calculated based on the energy-response function S0(CE). ref ); S5: A series of discrete collision energy values ​​CE1 to CE that are exactly the same as S2. n Below, the energy values ​​(CE) of the unknown concentration sample at each collision are measured. i The signal intensity I of the same specific mother ion-daughter ion pair generated below x (CE i ), and its energy at the reference collision energy value CE was measured. ref Signal strength I x (CE ref ), calculate the experimental response ratio R x (CE i )=I x (CE i ) / I x (CE ref ); S6: By minimizing Σ[R(CE) i )-R x (CE i )] 2 The concentration C of the target analyte in the sample with unknown concentration was obtained by inversion calculation. x .

2. The energy-dilution-based mass spectrometry detection and quantitative method as described in claim 1, characterized in that, The interval between adjacent collision energy values ​​is less than 0.5 eV, and the collision energy values ​​CE1 to CE n It is located in the range of 0-80 eV.

3. The energy dilution-based mass spectrometry detection and quantitative method as described in claim 1, characterized in that, In S3, based on the discrete dataset {CE i , S0(CE i Identify the energy values ​​CE at two inflection points. a and CE b , of which CE a and CE b Meets CE1 <CE a <CE b <CE n ; The continuous range of collision energy is divided into three sub-regions: a low-energy region, a transition region, and a high-energy region. The low-energy region is from CE1 to less than CE. a The energy range is fitted using a logistic function model; the transition region is from CE. a To CE b The energy range was fitted using a polynomial function model; the high-energy range was from greater than CE. b To CE n The energy range was fitted using an exponential decay function model to construct a model covering CE1 to CE2. n The continuous energy-response function S0(CE).

4. The energy dilution-based mass spectrometry detection and quantification method as described in claim 3, characterized in that, The region connecting the low-energy region and the transition region is defined as a closed interval [CE]. a -ΔE, CE a +ΔE], where ΔE is the transition bandwidth, calculated using the formula ΔE=0.15×(CE). b -CE a Calculations show that, within the connected region, the energy-response function S0(CE) is expressed as S0(CE) = f low (CE)×[1-T(CE)]+f mid (CE)×T(CE), where, T(CE)=[1+tanh((CE-CE a ) / ΔE)] / 2; The region connecting the transition region and the high-energy region is defined as a closed interval [CE]. b -ΔE, CE b +ΔE], within the connected region, the energy-response function S0(CE) is expressed as: S0(CE) = f mid (CE)×[1-U(CE)]+f high (CE)×U(CE), where U(CE)=[1+tanh((CE-CE b ) / ΔE)] / 2; Among them, f low (CE), f mid (CE), f high (CE) represent the logistic function model, the polynomial function model, and the exponential decay function model, respectively.

5. The energy-dilution-based mass spectrometry detection and quantification method as described in claim 1, characterized in that, The minimum dissociation energy barrier E of the parent ion of the target analyte was determined in advance through quantum chemical calculations. diss and the main transition state energy ETS; for discrete datasets {CE i ,S0(CE i The smoothing process is performed to generate a continuous response curve. The first derivative dS0 / d(CE) and the second derivative dS0 / d(CE) of the continuous response curve are then calculated. 2 S0 / d(CE) 2 ; Local maxima of the second derivative are defined as candidate inflection points. Candidate inflection points satisfying the following conditions are selected as valid inflection points: at the candidate inflection point, the absolute value of the first derivative is greater than a preset threshold δ; the signs of the second derivatives on either side of the candidate inflection point are opposite; |CE candidate -Ediss|≤0.15×E diss or |CE candidate -ETS|≤0.15×ETS, CE candidate The energy value corresponding to the candidate inflection point; Sort all valid inflection points in ascending order of energy value, and denote the energy value corresponding to the valid inflection point with the smallest energy value as CE. a The energy value corresponding to the effective inflection point with the largest energy value is denoted as CE. b And meets CE requirements a <CE b .

6. The energy-dilution-based mass spectrometry detection and quantification method as described in claim 1, characterized in that, In S6, the concentration C is calculated by inversion. x The specific steps include: Construct the objective function F(C) x )=Σ[R(CE i )–R x (CE i )] 2 ; Initialize concentration iteration value C x(0) C x(0) ∈(0, C0]; According to the iterative formula C x(k+1) =C x(k) -γ·▽F(C x(k) Update the concentration value, where k is the iteration number, γ is the step size factor, and ▽F is F(C x The gradient of ). When |C x(k+1) -C x(k) |≤ε or F (Cx(k)) Iteration stops when η ≤ η, and outputs C. x =C x(k+1) , where ε is the concentration convergence threshold and η is the objective function convergence threshold.

7. The energy-dilution-based mass spectrometry detection and quantification method as described in claim 6, characterized in that, The step size factor γ employs an adaptive adjustment strategy, including: Initialize step size γ (0) =0.01×C0; Starting from the k=1th iteration, according to γ (k) =|(C x(k) -C x(k-1) )·(▽F (k) -▽F (k-1) )| / ||▽F (k) -▽F (k-1) || 2 , where γ (k) C represents the step size factor for the k-th iteration. x(k) Let F represent the concentration estimate in the k-th iteration. (k) The objective function F(C) represents the objective function. x In C x = C x(k) The gradient value at that point; Constraint updated step size γ (k) Within the range [0.001×C0, 0.1×C0].

8. The energy-dilution-based mass spectrometry detection and quantification method as described in claim 5, characterized in that, Also includes: Calculate the objective function f(C) x Curvature parameter H, ; Calculate the concentration variance Var(C) x ): , σ I σ is the standard measurement uncertainty of the mass spectrometry signal intensity. I =0.02×I x (CE ref ) Calculate the standard uncertainty u(C) x ), ; Calculate the relative expanded uncertainty U rel , , where k is the inclusion factor.

9. An apparatus for implementing the energy dilution-based mass spectrometry detection and quantitative method as described in claim 1, characterized in that, include: Energy scan sequence generator, used to generate collision energy value sequences CE1 to CE n ; The energy-response function fitting module is used to receive standard solutions at CE1 to CE2. n Signal strength I0 (CE) i ), calculate S0(CE) i )=I0(CE i S0(CE) is a continuous function fitted using a machine learning algorithm. The drift correction and concentration inversion module is used to select CE. ref And calculate R(CE) = S0(CE) / S0(CE) ref Based on unknown sample signal I x (CE i ) and I x (CE ref ), calculate R x (CE i )=I x (CE i ) / I x (CE ref ); by minimizing Σ[R(CE) i )–R x (CE i )] 2 Inversion calculation to solve C x ; The uncertainty assessment module is used to output the standard uncertainty and relative expanded uncertainty of the concentration Cx.

Images